import pandas as pd
import numpy as np
import math
import seaborn as sns
#from numpy import array, hstack, math
#from numpy.random import uniform
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split,cross_val_score,RepeatedKFold
from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor,GradientBoostingRegressor
from sklearn.metrics import mean_squared_error, r2_score, explained_variance_score, mean_absolute_error
from sklearn.feature_selection import RFE
from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import StandardScaler
from sklearn.neighbors import KNeighborsRegressor
from sklearn import set_config
from sklearn.svm import LinearSVR
from sklearn.multioutput import MultiOutputRegressor,RegressorChain
import lightgbm as lgb


# Make numpy printouts easier to read.
np.set_printoptions(precision=3, suppress=True)
pd.set_option('display.max_columns', None)
pd.set_option('precision', 2)

#Read the csv file
df=pd.read_csv('C:/Users/ravip/Desktop/Florida_Orlando_Data.csv' , skiprows=9)
print(df.head(5))

       timestamp  Temp_2m  RH_2m  MSL_Pressure  Precipitation_Total  Snowfall  \
0  20000101T0000    18.45     73        1021.9                  0.0         0   
1  20000101T0100    18.02     75        1022.0                  0.0         0   
2  20000101T0200    17.55     77        1021.9                  0.0         0   
3  20000101T0300    17.03     80        1021.5                  0.0         0   
4  20000101T0400    16.57     82        1021.5                  0.0         0   

   total_cld_lay  high_cld_lay  mid_cld_lay  low_cld_lay  Sunshine_Duration  \
0           18.0             0            0           18                0.0   
1           20.0             0            0           20                0.0   
2           21.0             0            0           21                0.0   
3           15.0             0            0           15                0.0   
4           25.0             0            0           25                0.0   

   Shortwave_Radiation  Direct_Shortwave_Radiation  \
0                  0.0                         0.0   
1                  0.0                         0.0   
2                  0.0                         0.0   
3                  0.0                         0.0   
4                  0.0                         0.0   

   Diffuse_Shortwave_Radiation  Evptrp  FAO_Ref_Evptrp_2m  CAPE  Wind_Gust  \
0                          0.0     0.0           2.77e-03    24       0.36   
1                          0.0     0.0           0.00e+00    18       1.08   
2                          0.0     0.0           0.00e+00     7       2.88   
3                          0.0     0.0           0.00e+00     0       4.68   
4                          0.0     0.0           0.00e+00     3       5.76   

   Wind_Speed_10m  Wind_Dir_10m  Wind_Speed_80m  Wind_Dir_80m  \
0            3.60        143.13            6.29        166.76   
1            3.10        144.46            6.83        161.57   
2            3.05        135.00            7.09        156.04   
3            3.10        125.54            7.10        149.53   
4            3.40        122.01            7.59        148.57   

   Wind_Speed_900mb  Wind_Dir_900mb  Wind_Speed_850mb  Wind_Dir_850mb  \
0             10.97          221.01             21.92          227.66   
1             12.54          219.17             23.11          232.59   
2             13.55          219.61             23.71          239.93   
3             15.04          222.09             24.54          247.57   
4             15.79          223.15             25.52          253.61   

   Wind_Speed_700_mb]  Wind_Dir_700mb  Wind_Speed_500mb  Wind_Dir_500mb  \
0               31.36          257.40             62.01          254.51   
1               30.19          261.77             61.67          254.42   
2               29.22          266.47             62.08          255.56   
3               28.44          270.73             62.48          258.37   
4               28.59          275.78             62.78          262.75   

   GP_Height_1000mb  GP_Height_850mb  GP_Height_700mb  GP_Height_500mb  \
0               185             1556             3144             5799   
1               186             1557             3144             5799   
2               185             1556             3143             5797   
3               182             1552             3138             5791   
4               182             1553             3138             5791   

   Temp_1000mb  Temp_850mb  Temp_700mb   Temp  Soil_Temp  Soil_Moisture  \
0        19.15        8.42        4.54  14.17      18.16           0.13   
1        18.88        8.56        4.35  13.30      16.96           0.13   
2        18.67        8.73        4.12  12.64      15.94           0.13   
3        18.52        8.88        3.82  12.12      15.08           0.13   
4        18.45        9.04        3.54  11.73      14.37           0.13   

   Vapor_Pressure_Deficit_2m  
0                       5.74  
1                       5.18  
2                       4.62  
3                       3.89  
4                       3.40  

Exploratory Analysis

#convert the timestamp object type variable into a date time variable to extract year, month, day and time. 
df['timestamp'] =  pd.to_datetime(df['timestamp'])
df['day'] = df['timestamp'].dt.day
df['year'] = df['timestamp'].dt.year
df['month'] = df['timestamp'].dt.month

def hr_func(ts):
    return ts.hour

df['hour_of_day'] = df['timestamp'].apply(hr_func)

#drop the columns that have no data or which are irrelevant to the study
#snowfall feature does not have any data since Orlando is a sub tropical location which does not get any snowfall. 
#time stamp feature has been extracted into multiple fields for the ease of analysis and hence this field will be dropped as it has string data that is no longer needed for processing. 
data = df.drop(['timestamp', 'Snowfall'], axis=1)
print(data.head(5))

   Temp_2m  RH_2m  MSL_Pressure  Precipitation_Total  total_cld_lay  \
0    18.45     73        1021.9                  0.0           18.0   
1    18.02     75        1022.0                  0.0           20.0   
2    17.55     77        1021.9                  0.0           21.0   
3    17.03     80        1021.5                  0.0           15.0   
4    16.57     82        1021.5                  0.0           25.0   

   high_cld_lay  mid_cld_lay  low_cld_lay  Sunshine_Duration  \
0             0            0           18                0.0   
1             0            0           20                0.0   
2             0            0           21                0.0   
3             0            0           15                0.0   
4             0            0           25                0.0   

   Shortwave_Radiation  Direct_Shortwave_Radiation  \
0                  0.0                         0.0   
1                  0.0                         0.0   
2                  0.0                         0.0   
3                  0.0                         0.0   
4                  0.0                         0.0   

   Diffuse_Shortwave_Radiation  Evptrp  FAO_Ref_Evptrp_2m  CAPE  Wind_Gust  \
0                          0.0     0.0           2.77e-03    24       0.36   
1                          0.0     0.0           0.00e+00    18       1.08   
2                          0.0     0.0           0.00e+00     7       2.88   
3                          0.0     0.0           0.00e+00     0       4.68   
4                          0.0     0.0           0.00e+00     3       5.76   

   Wind_Speed_10m  Wind_Dir_10m  Wind_Speed_80m  Wind_Dir_80m  \
0            3.60        143.13            6.29        166.76   
1            3.10        144.46            6.83        161.57   
2            3.05        135.00            7.09        156.04   
3            3.10        125.54            7.10        149.53   
4            3.40        122.01            7.59        148.57   

   Wind_Speed_900mb  Wind_Dir_900mb  Wind_Speed_850mb  Wind_Dir_850mb  \
0             10.97          221.01             21.92          227.66   
1             12.54          219.17             23.11          232.59   
2             13.55          219.61             23.71          239.93   
3             15.04          222.09             24.54          247.57   
4             15.79          223.15             25.52          253.61   

   Wind_Speed_700_mb]  Wind_Dir_700mb  Wind_Speed_500mb  Wind_Dir_500mb  \
0               31.36          257.40             62.01          254.51   
1               30.19          261.77             61.67          254.42   
2               29.22          266.47             62.08          255.56   
3               28.44          270.73             62.48          258.37   
4               28.59          275.78             62.78          262.75   

   GP_Height_1000mb  GP_Height_850mb  GP_Height_700mb  GP_Height_500mb  \
0               185             1556             3144             5799   
1               186             1557             3144             5799   
2               185             1556             3143             5797   
3               182             1552             3138             5791   
4               182             1553             3138             5791   

   Temp_1000mb  Temp_850mb  Temp_700mb   Temp  Soil_Temp  Soil_Moisture  \
0        19.15        8.42        4.54  14.17      18.16           0.13   
1        18.88        8.56        4.35  13.30      16.96           0.13   
2        18.67        8.73        4.12  12.64      15.94           0.13   
3        18.52        8.88        3.82  12.12      15.08           0.13   
4        18.45        9.04        3.54  11.73      14.37           0.13   

   Vapor_Pressure_Deficit_2m  day  year  month  hour_of_day  
0                       5.74    1  2000      1            0  
1                       5.18    1  2000      1            1  
2                       4.62    1  2000      1            2  
3                       3.89    1  2000      1            3  
4                       3.40    1  2000      1            4  

#check for the info regarding the data
data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 182448 entries, 0 to 182447
Data columns (total 43 columns):
 #   Column                       Non-Null Count   Dtype  
---  ------                       --------------   -----  
 0   Temp_2m                      182448 non-null  float64
 1   RH_2m                        182448 non-null  int64  
 2   MSL_Pressure                 182448 non-null  float64
 3   Precipitation_Total          182448 non-null  float64
 4   total_cld_lay                182448 non-null  float64
 5   high_cld_lay                 182448 non-null  int64  
 6   mid_cld_lay                  182448 non-null  int64  
 7   low_cld_lay                  182448 non-null  int64  
 8   Sunshine_Duration            182448 non-null  float64
 9   Shortwave_Radiation          182448 non-null  float64
 10  Direct_Shortwave_Radiation   182448 non-null  float64
 11  Diffuse_Shortwave_Radiation  182448 non-null  float64
 12  Evptrp                       182448 non-null  float64
 13  FAO_Ref_Evptrp_2m            182448 non-null  float64
 14  CAPE                         182448 non-null  int64  
 15  Wind_Gust                    182448 non-null  float64
 16  Wind_Speed_10m               182448 non-null  float64
 17  Wind_Dir_10m                 182448 non-null  float64
 18  Wind_Speed_80m               182448 non-null  float64
 19  Wind_Dir_80m                 182448 non-null  float64
 20  Wind_Speed_900mb             182448 non-null  float64
 21  Wind_Dir_900mb               182448 non-null  float64
 22  Wind_Speed_850mb             182448 non-null  float64
 23  Wind_Dir_850mb               182448 non-null  float64
 24  Wind_Speed_700_mb]           182448 non-null  float64
 25  Wind_Dir_700mb               182448 non-null  float64
 26  Wind_Speed_500mb             182448 non-null  float64
 27  Wind_Dir_500mb               182448 non-null  float64
 28  GP_Height_1000mb             182448 non-null  int64  
 29  GP_Height_850mb              182448 non-null  int64  
 30  GP_Height_700mb              182448 non-null  int64  
 31  GP_Height_500mb              182448 non-null  int64  
 32  Temp_1000mb                  182448 non-null  float64
 33  Temp_850mb                   182448 non-null  float64
 34  Temp_700mb                   182448 non-null  float64
 35  Temp                         182448 non-null  float64
 36  Soil_Temp                    182448 non-null  float64
 37  Soil_Moisture                182448 non-null  float64
 38  Vapor_Pressure_Deficit_2m    182448 non-null  float64
 39  day                          182448 non-null  int64  
 40  year                         182448 non-null  int64  
 41  month                        182448 non-null  int64  
 42  hour_of_day                  182448 non-null  int64  
dtypes: float64(30), int64(13)
memory usage: 59.9 MB

#check for any missing or null values in the features and labels. 
data.isnull().sum()

Temp_2m                        0
RH_2m                          0
MSL_Pressure                   0
Precipitation_Total            0
total_cld_lay                  0
high_cld_lay                   0
mid_cld_lay                    0
low_cld_lay                    0
Sunshine_Duration              0
Shortwave_Radiation            0
Direct_Shortwave_Radiation     0
Diffuse_Shortwave_Radiation    0
Evptrp                         0
FAO_Ref_Evptrp_2m              0
CAPE                           0
Wind_Gust                      0
Wind_Speed_10m                 0
Wind_Dir_10m                   0
Wind_Speed_80m                 0
Wind_Dir_80m                   0
Wind_Speed_900mb               0
Wind_Dir_900mb                 0
Wind_Speed_850mb               0
Wind_Dir_850mb                 0
Wind_Speed_700_mb]             0
Wind_Dir_700mb                 0
Wind_Speed_500mb               0
Wind_Dir_500mb                 0
GP_Height_1000mb               0
GP_Height_850mb                0
GP_Height_700mb                0
GP_Height_500mb                0
Temp_1000mb                    0
Temp_850mb                     0
Temp_700mb                     0
Temp                           0
Soil_Temp                      0
Soil_Moisture                  0
Vapor_Pressure_Deficit_2m      0
day                            0
year                           0
month                          0
hour_of_day                    0
dtype: int64

Upon examining no categorical data is observed in any of the columns. Also, none of the columns seems to have any missing values.

#drop any duplicate records from the data
data = data.drop_duplicates()
data.shape

(182448, 43)

It can be seen that the data has well balanced row vs. column proportions.

data.describe()

	Temp_2m	RH_2m	MSL_Pressure	Precipitation_Total	total_cld_lay	high_cld_lay	mid_cld_lay	low_cld_lay	Sunshine_Duration	Shortwave_Radiation	Direct_Shortwave_Radiation	Diffuse_Shortwave_Radiation	Evptrp	FAO_Ref_Evptrp_2m	CAPE	Wind_Gust	Wind_Speed_10m	Wind_Dir_10m	Wind_Speed_80m	Wind_Dir_80m	Wind_Speed_900mb	Wind_Dir_900mb	Wind_Speed_850mb	Wind_Dir_850mb	Wind_Speed_700_mb]	Wind_Dir_700mb	Wind_Speed_500mb	Wind_Dir_500mb	GP_Height_1000mb	GP_Height_850mb	GP_Height_700mb	GP_Height_500mb	Temp_1000mb	Temp_850mb	Temp_700mb	Temp	Soil_Temp	Soil_Moisture	Vapor_Pressure_Deficit_2m	day	year	month	hour_of_day
count	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	1.82e+05	1.82e+05	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	182448.00	1.82e+05	182448.00	182448.00	182448.00	182448.00
mean	22.13	72.87	1017.76	0.07	41.77	25.71	24.06	28.87	16.81	182.82	105.68	77.14	9.59e-02	1.48e-01	327.79	18.44	12.27	160.23	18.54	162.79	23.08	180.52	23.39	194.10	29.88	217.07	45.21	234.73	151.17	1541.25	3154.69	5833.44	21.83	13.35	6.11	22.94	22.75	0.16	8.41e+00	15.72	2009.91	6.48	11.50
std	6.28	17.89	4.37	0.38	42.98	39.56	39.70	39.69	23.97	252.53	171.48	96.07	1.27e-01	1.85e-01	418.77	8.85	6.67	95.91	8.14	95.12	15.07	91.57	15.71	93.05	21.28	88.86	33.49	80.82	35.51	33.16	43.07	65.12	5.58	3.92	2.90	7.70	5.94	0.03	7.28e+00	8.80	6.01	3.43	6.92
min	-6.25	14.00	975.60	0.00	0.00	0.00	0.00	0.00	0.00	0.00	-13.46	0.00	0.00e+00	0.00e+00	0.00	0.00	0.00	0.60	0.00	0.56	0.00	0.28	0.00	0.31	0.00	0.34	0.00	0.21	-207.00	1202.00	2866.00	5409.00	-6.79	-7.68	-12.75	-6.56	0.58	0.04	-1.91e-06	1.00	2000.00	1.00	0.00
25%	18.45	59.00	1014.90	0.00	0.90	0.00	0.00	0.00	0.00	0.00	0.00	0.00	1.44e-03	5.41e-03	0.00	11.88	7.29	81.87	12.74	85.07	11.97	100.84	12.02	112.11	14.34	163.30	19.32	217.18	129.00	1523.00	3132.00	5795.00	18.65	11.16	4.33	18.42	19.18	0.15	2.48e+00	8.00	2005.00	4.00	5.75
50%	23.04	75.00	1017.50	0.00	16.20	0.00	0.00	10.00	0.00	15.13	2.05	6.86	1.87e-02	5.07e-02	122.00	18.00	10.83	138.01	18.71	139.30	20.09	185.03	20.07	213.69	24.27	243.43	35.65	257.47	150.00	1545.00	3161.00	5846.00	22.92	14.15	6.70	23.02	23.61	0.17	6.30e+00	16.00	2010.00	6.00	11.50
75%	26.66	89.00	1020.40	0.00	100.00	41.00	48.00	67.00	43.15	316.84	157.47	158.89	1.67e-01	2.51e-01	562.00	23.40	16.20	243.43	23.77	248.96	30.86	254.17	30.93	266.19	40.16	279.21	64.67	283.39	173.00	1564.00	3185.00	5884.00	25.89	16.48	8.29	28.01	26.86	0.18	1.27e+01	23.00	2015.00	9.00	17.25
max	39.00	100.00	1037.10	27.60	100.00	100.00	100.00	100.00	60.00	996.80	690.80	340.88	6.15e-01	9.10e-01	4277.00	104.40	78.41	360.00	104.52	360.00	173.22	360.00	182.76	360.00	176.24	360.00	228.53	360.00	298.00	1651.00	3268.00	5970.00	36.30	22.08	17.81	48.80	39.42	0.30	5.47e+01	31.00	2020.00	12.00	23.00

sns.set(rc={'figure.figsize':(17,17)})
# Calculate correlations
correlations = data.corr()
 
# Print the Correlation Heatmap using seaborn
sns.heatmap(correlations)

<AxesSubplot:>

Creation and Analysis of Features and Targets

# Dividing dataset into label and feature sets
#features = data_modified.drop(['Temp_2m','RH_2m'], axis = 1) # Features
#labels = data_modified[['Temp_2m','RH_2m']] # Labels
features = data.drop(['Temp_2m','RH_2m'], axis = 1) # Features
labels = data[['Temp_2m','RH_2m']] # Labels
print(type(features))
print(type(labels))
print(features.shape)
print(labels.shape)

<class 'pandas.core.frame.DataFrame'>
<class 'pandas.core.frame.DataFrame'>
(182448, 41)
(182448, 2)

# feature importance estimation to see which features can help in better training of the model. 
feature_importance = DecisionTreeRegressor()
feature_importance.fit(features,labels)
print(feature_importance.feature_importances_)
for feat, importance in zip(features.columns, feature_importance.feature_importances_):
    print ('feature: {f}, importance: {i}'.format(f=feat, i=importance))

[0.    0.    0.    0.    0.    0.    0.    0.    0.    0.    0.    0.
 0.032 0.    0.    0.    0.    0.    0.    0.    0.    0.    0.    0.
 0.    0.    0.    0.    0.    0.    0.123 0.008 0.    0.041 0.001 0.
 0.791 0.    0.    0.001 0.   ]
feature: MSL_Pressure, importance: 5.9178574476959616e-05
feature: Precipitation_Total, importance: 4.97035440399459e-06
feature: total_cld_lay, importance: 2.1617564243998973e-05
feature: high_cld_lay, importance: 1.2584614786614461e-05
feature: mid_cld_lay, importance: 8.72101021209456e-06
feature: low_cld_lay, importance: 4.603057530062024e-05
feature: Sunshine_Duration, importance: 4.1262250202964235e-05
feature: Shortwave_Radiation, importance: 5.055077511726679e-05
feature: Direct_Shortwave_Radiation, importance: 7.946729741094226e-05
feature: Diffuse_Shortwave_Radiation, importance: 6.044691321479496e-05
feature: Evptrp, importance: 7.510901172714087e-05
feature: FAO_Ref_Evptrp_2m, importance: 0.00033593825538319835
feature: CAPE, importance: 0.03195203888317169
feature: Wind_Gust, importance: 8.925587468961249e-05
feature: Wind_Speed_10m, importance: 9.313587077140822e-05
feature: Wind_Dir_10m, importance: 6.275760355783387e-05
feature: Wind_Speed_80m, importance: 0.00011505251666989408
feature: Wind_Dir_80m, importance: 7.292324593609635e-05
feature: Wind_Speed_900mb, importance: 6.133734828924194e-05
feature: Wind_Dir_900mb, importance: 7.542159873001992e-05
feature: Wind_Speed_850mb, importance: 5.745048623413823e-05
feature: Wind_Dir_850mb, importance: 6.785887433323778e-05
feature: Wind_Speed_700_mb], importance: 7.47255042725375e-05
feature: Wind_Dir_700mb, importance: 6.1519491770601e-05
feature: Wind_Speed_500mb, importance: 7.702519217232532e-05
feature: Wind_Dir_500mb, importance: 6.097164255501906e-05
feature: GP_Height_1000mb, importance: 6.787157428717148e-05
feature: GP_Height_850mb, importance: 9.765918256131429e-05
feature: GP_Height_700mb, importance: 0.00010582403696931768
feature: GP_Height_500mb, importance: 0.00034854497329230293
feature: Temp_1000mb, importance: 0.12312103474001601
feature: Temp_850mb, importance: 0.008096429456201527
feature: Temp_700mb, importance: 0.0004813106283136684
feature: Temp, importance: 0.04065489481891291
feature: Soil_Temp, importance: 0.0012342726773117902
feature: Soil_Moisture, importance: 0.00024056637395156551
feature: Vapor_Pressure_Deficit_2m, importance: 0.7907300616659558
feature: day, importance: 6.334186764938379e-05
feature: year, importance: 6.298015763046619e-05
feature: month, importance: 0.0009246175262858175
feature: hour_of_day, importance: 5.3238991026877485e-05

#split the data for training and testing
trainf,testf,trainl,testl=train_test_split(features,labels, test_size=0.15, random_state=22)
print(len(trainf))
print(len(trainl))
print(len(testf))
print(len(testl))
print("trainf:", trainf.shape, "trianl:", trainl.shape)
print("testf:", testf.shape, "testl:", testl.shape)

#Conver the pandas dataframes into arrays for train and test features
trainf_array = trainf.to_numpy()
trainl_array = trainl.to_numpy()
testf_array = testf.to_numpy()
testl_array = testl.to_numpy()
print("trainf_array:", trainf_array.shape, "trianl_array:", trainl_array.shape)
print("testf_array:", testf_array.shape, "testl_array:", testl_array.shape)

155080
155080
27368
27368
trainf: (155080, 41) trianl: (155080, 2)
testf: (27368, 41) testl: (27368, 2)
trainf_array: (155080, 41) trianl_array: (155080, 2)
testf_array: (27368, 41) testl_array: (27368, 2)

#Apply normalization on the training and testing feature sets 
scaler = StandardScaler()
# transform data
trainf_array_scaled = scaler.fit_transform(trainf_array)
testf_array_scaled = scaler.fit_transform(testf_array)

Linear Regression

set_config(print_changed_only=False)
model_LR = LinearRegression(normalize=True)

model_LR.fit(trainf_array,trainl_array)
LRpred = model_LR.predict(testf_array)
print(r2_score(LRpred, testl_array))

0.9575483230709314

The R2 accuracy is very high which means the features and targets are linearly dependent.

#Model Evaluation

mse_temp = mean_squared_error(testl_array[:,0], LRpred[:,0])
print("Mean Squared Error for Temperature is : ", mse_temp)

mse_humidity = mean_squared_error(testl_array[:,1], LRpred[:,1])
print(" Mean Squared Error for Relative Humidity is : ", mse_humidity)

rmse_temp = math.sqrt(mean_squared_error(testl_array[:,0], LRpred[:,0]))
print("Root Mean Squared Error for Temperature is : ", rmse_temp)

rmse_humidity = math.sqrt(mean_squared_error(testl_array[:,1], LRpred[:,1]))
print("Root Mean Squared Error for Relative Humidity is : ", rmse_humidity)

exp_var_score_temp = explained_variance_score(testl_array[:,0], LRpred[:,0])
print("The Explained variance score Temperature is {}".format(exp_var_score_temp))

exp_var_score_humidity = explained_variance_score(testl_array[:,1], LRpred[:,1])
print("The Explained variance score in terms of Relative Humidity is {}".format(exp_var_score_humidity))

mae_temp = mean_absolute_error(testl_array[:,0], LRpred[:,0])
print("Explained Variance Score for Temperature is ", mae_temp)

mae_hum = mean_absolute_error(testl_array[:,1], LRpred[:,1])
print("Explained Variance Score for Relative Humidity is ", mae_hum)

def mean_absolute_percentage_error(y_true, y_pred): 
    y_true, y_pred = np.array(y_true), np.array(y_pred)
    return np.mean(np.abs((y_true - y_pred) / (y_true))) * 100

mean_abs_perc_error_temp = mean_absolute_percentage_error(testl_array[:,0], LRpred[:,0])
print("The mean absolute percentage error for Temperature is" ,mean_abs_perc_error_temp)

mean_abs_perc_error_humidity = mean_absolute_percentage_error(testl_array[:,1], LRpred[:,1])
print("The mean absolute percentage error for Relative Humidity is" ,mean_abs_perc_error_humidity)

Mean Squared Error for Temperature is :  0.2053901206852193
 Mean Squared Error for Relative Humidity is :  23.611172413791728
Root Mean Squared Error for Temperature is :  0.4531998683640798
Root Mean Squared Error for Relative Humidity is :  4.85913288702745
The Explained variance score Temperature is 0.9948201836109379
The Explained variance score in terms of Relative Humidity is 0.9259200468779117
Explained Variance Score for Temperature is  0.3312862931946686
Explained Variance Score for Relative Humidity is  3.348559790511139
The mean absolute percentage error for Temperature is 3.2971421473688083
The mean absolute percentage error for Relative Humidity is 5.456512698160341

#Create a pandas dataframe to store the metrics of different models performed. 
ModelMetrics = pd.DataFrame()
ModelMetrics= ModelMetrics.append([['Linear Regression',mse_temp,mse_humidity,rmse_temp,rmse_humidity,exp_var_score_temp,exp_var_score_humidity,mae_temp,mae_hum,mean_abs_perc_error_temp,mean_abs_perc_error_humidity]])
#ModelMetrics.columns= ['ModelName','MSE-Temp','MSE-RH','RMSE-Temp','RMSE-RH','R2Score-Temp','R2Score-RH','MAE-Temp','MAE-RH','MAPE-Temp','MAPE-RH']

ModelMetrics

	0	1	2	3	4	5	6	7	8	9	10
0	Linear Regression	0.21	23.61	0.45	4.86	0.99	0.93	0.33	3.35	3.3	5.46

#Temperature Prediction
plt.plot(range(0,100), testl_array[0:100,0], label="Actual_Temp", color='r')
plt.scatter(range(0,100), LRpred[0:100,0], label="Predicted_Temp", color='b')
plt.xlabel('Index')
plt.ylabel('Temperature in Degrees Celcius')
plt.legend()
plt.show()

#Humidity Prediction
plt.plot(range(0,100), testl_array[0:100,1], label="Actual_Humidity", color='g')
plt.scatter(range(0,100), LRpred[0:100,1], label="Predicted_Humidity", color='y')
plt.xlabel('Index')
plt.ylabel('Percentage Humidity')
plt.legend()
plt.show()

#Temperature Scatter Plot for predictions
plt.scatter(testl_array[:,0], LRpred[:,0])
plt.ylabel("Predicted Temperatures")
plt.xlabel("Test Temperatures")
plt.title(" Predicted Vs. Test Temperature")
plt.show()

#RH Scatter Plot for predictions
plt.scatter(testl_array[:,1], LRpred[:,1])
plt.ylabel("Predicted RH")
plt.xlabel("Test RH")
plt.title(" Predicted Vs. Test RH")
plt.show()

KNN Regression

model_KNN = KNeighborsRegressor()
model_KNN.fit(trainf_array_scaled,trainl_array)

KNeighborsRegressor(algorithm='auto', leaf_size=30, metric='minkowski',
                    metric_params=None, n_jobs=None, n_neighbors=5, p=2,
                    weights='uniform')

KNNpred = model_KNN.predict(testf_array_scaled)

#Model Evaluation

mse_temp = mean_squared_error(testl_array[:,0], KNNpred[:,0])
print("Mean Squared Error for Temperature is : ", mse_temp)

mse_humidity = mean_squared_error(testl_array[:,1], KNNpred[:,1])
print(" Mean Squared Error for Relative Humidity is : ", mse_humidity)

rmse_temp = math.sqrt(mean_squared_error(testl_array[:,0], KNNpred[:,0]))
print("Root Mean Squared Error for Temperature is : ", rmse_temp)

rmse_humidity = math.sqrt(mean_squared_error(testl_array[:,1], KNNpred[:,1]))
print("Root Mean Squared Error for Relative Humidity is : ", rmse_humidity)

exp_var_score_temp = explained_variance_score(testl_array[:,0], KNNpred[:,0])
print("The Explained variance score Temperature is {}".format(exp_var_score_temp))

exp_var_score_humidity = explained_variance_score(testl_array[:,1], KNNpred[:,1])
print("The Explained variance score in terms of Relative Humidity is {}".format(exp_var_score_humidity))

mae_temp = mean_absolute_error(testl_array[:,0], KNNpred[:,0])
print("Explained Variance Score for Temperature is ", mae_temp)

mae_hum = mean_absolute_error(testl_array[:,1], KNNpred[:,1])
print("Explained Variance Score for Relative Humidity is ", mae_hum)

mean_abs_perc_error_temp = mean_absolute_percentage_error(testl_array[:,0], KNNpred[:,0])
print("The mean absolute percentage error for Temperature is" ,mean_abs_perc_error_temp)

mean_abs_perc_error_humidity = mean_absolute_percentage_error(testl_array[:,1], KNNpred[:,1])
print("The mean absolute percentage error for Relative Humidity is" ,mean_abs_perc_error_humidity)

Mean Squared Error for Temperature is :  0.8755923556772365
 Mean Squared Error for Relative Humidity is :  21.92111663256358
Root Mean Squared Error for Temperature is :  0.9357309205520765
Root Mean Squared Error for Relative Humidity is :  4.681999213216891
The Explained variance score Temperature is 0.9779243848328526
The Explained variance score in terms of Relative Humidity is 0.9313580480958321
Explained Variance Score for Temperature is  0.6861658010998319
Explained Variance Score for Relative Humidity is  3.3315697164571763
The mean absolute percentage error for Temperature is 5.813117037860256
The mean absolute percentage error for Relative Humidity is 5.079609244263588

ModelMetrics= ModelMetrics.append([['KNN Regression',mse_temp,mse_humidity,rmse_temp,rmse_humidity,exp_var_score_temp,exp_var_score_humidity,mae_temp,mae_hum,mean_abs_perc_error_temp,mean_abs_perc_error_humidity]])

ModelMetrics

	0	1	2	3	4	5	6	7	8	9	10
0	Linear Regression	0.21	23.61	0.45	4.86	0.99	0.93	0.33	3.35	3.30	5.46
0	KNN Regression	0.88	21.92	0.94	4.68	0.98	0.93	0.69	3.33	5.81	5.08

#Temperature Prediction
plt.plot(range(0,100), testl_array[0:100,0], label="Actual_Temp", color='r')
plt.scatter(range(0,100), KNNpred[0:100,0], label="Predicted_Temp", color='b')
plt.xlabel('Index')
plt.ylabel('Temperature in Degrees Celcius')
plt.legend()
plt.show()

#RH Prediction
plt.plot(range(0,100), testl_array[0:100,1], label="Actual_Humidity", color='g')
plt.scatter(range(0,100), KNNpred[0:100,1], label="Predicted_Humidity", color='y')
plt.xlabel('Index')
plt.ylabel('Percentage Humidity')
plt.legend()
plt.show()

#Temperature Scatter Plot for predictions
plt.scatter(testl_array[:,0], KNNpred[:,0])
plt.ylabel("Predicted Temperatures")
plt.xlabel("Test Temperatures")
plt.title(" Predicted Vs. Test Temperature")
plt.show()

#RH Scatter Plot for predictions
plt.scatter(testl_array[:,1], KNNpred[:,1])
plt.ylabel("Predicted RH")
plt.xlabel("Test RH")
plt.title(" Predicted Vs. Test RH")
plt.show()

Support Vector Regression : Using Multi-Output Regressor

model_SVR = LinearSVR()
wrapper_SVR = MultiOutputRegressor(model_SVR)
wrapper_SVR.fit(trainf_array_scaled,trainl_array)

C:\Users\ravip\anaconda3\lib\site-packages\sklearn\svm\_base.py:976: ConvergenceWarning: Liblinear failed to converge, increase the number of iterations.
  warnings.warn("Liblinear failed to converge, increase "

MultiOutputRegressor(estimator=LinearSVR(C=1.0, dual=True, epsilon=0.0,
                                         fit_intercept=True,
                                         intercept_scaling=1.0,
                                         loss='epsilon_insensitive',
                                         max_iter=1000, random_state=None,
                                         tol=0.0001, verbose=0),
                     n_jobs=None)

SVRpred = wrapper_SVR.predict(testf_array_scaled)

#Model Evaluation

mse_temp = mean_squared_error(testl_array[:,0], SVRpred[:,0])
print("Mean Squared Error for Temperature is : ", mse_temp)

mse_humidity = mean_squared_error(testl_array[:,1], SVRpred[:,1])
print("Mean Squared Error for Relative Humidity is : ", mse_humidity)

rmse_temp = math.sqrt(mean_squared_error(testl_array[:,0], SVRpred[:,0]))
print("Root Mean Squared Error for Temperature is : ", rmse_temp)

rmse_humidity = math.sqrt(mean_squared_error(testl_array[:,1], SVRpred[:,1]))
print("Root Mean Squared Error for Relative Humidity is : ", rmse_humidity)

exp_var_score_temp = explained_variance_score(testl_array[:,0], SVRpred[:,0])
print("The Explained variance score Temperature is {}".format(exp_var_score_temp))

exp_var_score_humidity = explained_variance_score(testl_array[:,1], SVRpred[:,1])
print("The Explained variance score in terms of Relative Humidity is {}".format(exp_var_score_humidity))

mae_temp = mean_absolute_error(testl_array[:,0], SVRpred[:,0])
print("Explained Variance Score for Temperature is ", mae_temp)

mae_hum = mean_absolute_error(testl_array[:,1], SVRpred[:,1])
print("Explained Variance Score for Relative Humidity is ", mae_hum)

mean_abs_perc_error_temp = mean_absolute_percentage_error(testl_array[:,0], SVRpred[:,0])
print("The mean absolute percentage error for Temperature is" ,mean_abs_perc_error_temp)

mean_abs_perc_error_humidity = mean_absolute_percentage_error(testl_array[:,1], SVRpred[:,1])
print("The mean absolute percentage error for Relative Humidity is" ,mean_abs_perc_error_humidity)

Mean Squared Error for Temperature is :  0.21813062283558748
Mean Squared Error for Relative Humidity is :  24.599737729617775
Root Mean Squared Error for Temperature is :  0.46704456193771005
Root Mean Squared Error for Relative Humidity is :  4.959812267578056
The Explained variance score Temperature is 0.9945436488807007
The Explained variance score in terms of Relative Humidity is 0.922915751950394
Explained Variance Score for Temperature is  0.3257398925780984
Explained Variance Score for Relative Humidity is  3.2621434549466137
The mean absolute percentage error for Temperature is 3.6084152629758677
The mean absolute percentage error for Relative Humidity is 5.504315743327771

ModelMetrics= ModelMetrics.append([['SVR-Multioutput',mse_temp,mse_humidity,rmse_temp,rmse_humidity,exp_var_score_temp,exp_var_score_humidity,mae_temp,mae_hum,mean_abs_perc_error_temp,mean_abs_perc_error_humidity]])

ModelMetrics

	0	1	2	3	4	5	6	7	8	9	10
0	Linear Regression	0.21	23.61	0.45	4.86	0.99	0.93	0.33	3.35	3.30	5.46
0	KNN Regression	0.88	21.92	0.94	4.68	0.98	0.93	0.69	3.33	5.81	5.08
0	SVR-Multioutput	0.22	24.60	0.47	4.96	0.99	0.92	0.33	3.26	3.61	5.50

#Temperature Prediction
plt.plot(range(0,100), testl_array[0:100,0], label="Actual_Temp", color='r')
plt.scatter(range(0,100), SVRpred[0:100,0], label="Predicted_Temp", color='b')
plt.xlabel('Index')
plt.ylabel('Temperature in Degrees Celcius')
plt.legend()
plt.show()

#RH Prediction
plt.plot(range(0,100), testl_array[0:100,1], label="Actual_Humidity", color='g')
plt.scatter(range(0,100), SVRpred[0:100,1], label="Predicted_Humidity", color='y')
plt.xlabel('Index')
plt.ylabel('Percentage Humidity')
plt.legend()
plt.show()

#Temperature Scatter Plot for predictions
plt.scatter(testl_array[:,0], SVRpred[:,0])
plt.ylabel("Predicted Temperatures")
plt.xlabel("Test Temperatures")
plt.title(" Predicted Vs. Test Temperature")
plt.show()

#RH Scatter Plot for predictions
plt.scatter(testl_array[:,1], SVRpred[:,1])
plt.ylabel("Predicted RH")
plt.xlabel("Test RH")
plt.title(" Predicted Vs. Test RH")
plt.show()

Support Vector Regression : Using Regressor Chain

wrapper_RCSVR=RegressorChain(model_SVR)
wrapper_RCSVR.fit(trainf_array_scaled,trainl_array)

C:\Users\ravip\anaconda3\lib\site-packages\sklearn\svm\_base.py:976: ConvergenceWarning: Liblinear failed to converge, increase the number of iterations.
  warnings.warn("Liblinear failed to converge, increase "
C:\Users\ravip\anaconda3\lib\site-packages\sklearn\svm\_base.py:976: ConvergenceWarning: Liblinear failed to converge, increase the number of iterations.
  warnings.warn("Liblinear failed to converge, increase "

RegressorChain(base_estimator=LinearSVR(C=1.0, dual=True, epsilon=0.0,
                                        fit_intercept=True,
                                        intercept_scaling=1.0,
                                        loss='epsilon_insensitive',
                                        max_iter=1000, random_state=None,
                                        tol=0.0001, verbose=0),
               cv=None, order=None, random_state=None)

RCSVRpred = wrapper_RCSVR.predict(testf_array_scaled)

#Model Evaluation

mse_temp = mean_squared_error(testl_array[:,0], RCSVRpred[:,0])
print("Mean Squared Error for Temperature is : ", mse_temp)

mse_humidity = mean_squared_error(testl_array[:,1], RCSVRpred[:,1])
print("Mean Squared Error for Relative Humidity is : ", mse_humidity)

rmse_temp = math.sqrt(mean_squared_error(testl_array[:,0], RCSVRpred[:,0]))
print("Root Mean Squared Error for Temperature is : ", rmse_temp)

rmse_humidity = math.sqrt(mean_squared_error(testl_array[:,1], RCSVRpred[:,1]))
print("Root Mean Squared Error for Relative Humidity is : ", rmse_humidity)

exp_var_score_temp = explained_variance_score(testl_array[:,0], RCSVRpred[:,0])
print("The Explained variance score Temperature is {}".format(exp_var_score_temp))

exp_var_score_humidity = explained_variance_score(testl_array[:,1], RCSVRpred[:,1])
print("The Explained variance score in terms of Relative Humidity is {}".format(exp_var_score_humidity))

mae_temp = mean_absolute_error(testl_array[:,0], RCSVRpred[:,0])
print("Explained Variance Score for Temperature is ", mae_temp)

mae_hum = mean_absolute_error(testl_array[:,1], RCSVRpred[:,1])
print("Explained Variance Score for Relative Humidity is ", mae_hum)

mean_abs_perc_error_temp = mean_absolute_percentage_error(testl_array[:,0], RCSVRpred[:,0])
print("The mean absolute percentage error for Temperature is" ,mean_abs_perc_error_temp)

mean_abs_perc_error_humidity = mean_absolute_percentage_error(testl_array[:,1], RCSVRpred[:,1])
print("The mean absolute percentage error for Relative Humidity is" ,mean_abs_perc_error_humidity)

Mean Squared Error for Temperature is :  0.2174440033688633
Mean Squared Error for Relative Humidity is :  24.643339368063298
Root Mean Squared Error for Temperature is :  0.46630891409972347
Root Mean Squared Error for Relative Humidity is :  4.964205814434298
The Explained variance score Temperature is 0.9945497698440212
The Explained variance score in terms of Relative Humidity is 0.9229126033007371
Explained Variance Score for Temperature is  0.32637323127021617
Explained Variance Score for Relative Humidity is  3.2724205740536103
The mean absolute percentage error for Temperature is 3.5828899511728864
The mean absolute percentage error for Relative Humidity is 5.487541557266725

ModelMetrics= ModelMetrics.append([['SVR-RegressorChain',mse_temp,mse_humidity,rmse_temp,rmse_humidity,exp_var_score_temp,exp_var_score_humidity,mae_temp,mae_hum,mean_abs_perc_error_temp,mean_abs_perc_error_humidity]])

ModelMetrics

	0	1	2	3	4	5	6	7	8	9	10
0	Linear Regression	0.21	23.61	0.45	4.86	0.99	0.93	0.33	3.35	3.30	5.46
0	KNN Regression	0.88	21.92	0.94	4.68	0.98	0.93	0.69	3.33	5.81	5.08
0	SVR-Multioutput	0.22	24.60	0.47	4.96	0.99	0.92	0.33	3.26	3.61	5.50
0	SVR-RegressorChain	0.22	24.64	0.47	4.96	0.99	0.92	0.33	3.27	3.58	5.49

#Temperature Prediction
plt.plot(range(0,100), testl_array[0:100,0], label="Actual_Temp", color='r')
plt.scatter(range(0,100), RCSVRpred[0:100,0], label="Predicted_Temp", color='b')
plt.xlabel('Index')
plt.ylabel('Temperature in Degrees Celcius')
plt.legend()
plt.show()

#RH Prediction
plt.plot(range(0,100), testl_array[0:100,1], label="Actual_Humidity", color='g')
plt.scatter(range(0,100), RCSVRpred[0:100,1], label="Predicted_Humidity", color='y')
plt.xlabel('Index')
plt.ylabel('Percentage Humidity')
plt.legend()
plt.show()

#Temperature Scatter Plot for predictions
plt.scatter(testl_array[:,0], RCSVRpred[:,0])
plt.ylabel("Predicted Temperatures")
plt.xlabel("Test Temperatures")
plt.title(" Predicted Vs. Test Temperature")
plt.show()

#RH Scatter Plot for predictions
plt.scatter(testl_array[:,1], RCSVRpred[:,1])
plt.ylabel("Predicted RH")
plt.xlabel("Test RH")
plt.title(" Predicted Vs. Test RH")
plt.show()

Decision Tree Regressor

model_DT = DecisionTreeRegressor()
model_DT.fit(trainf_array,trainl_array)

DecisionTreeRegressor(ccp_alpha=0.0, criterion='mse', max_depth=None,
                      max_features=None, max_leaf_nodes=None,
                      min_impurity_decrease=0.0, min_impurity_split=None,
                      min_samples_leaf=1, min_samples_split=2,
                      min_weight_fraction_leaf=0.0, presort='deprecated',
                      random_state=None, splitter='best')

DTpred = model_DT.predict(testf_array)

#Model Evaluation

mse_temp = mean_squared_error(testl_array[:,0], DTpred[:,0])
print("Mean Squared Error for Temperature is : ", mse_temp)

mse_humidity = mean_squared_error(testl_array[:,1], DTpred[:,1])
print("Mean Squared Error for Relative Humidity is : ", mse_humidity)

rmse_temp = math.sqrt(mean_squared_error(testl_array[:,0], DTpred[:,0]))
print("Root Mean Squared Error for Temperature is : ", rmse_temp)

rmse_humidity = math.sqrt(mean_squared_error(testl_array[:,1], DTpred[:,1]))
print("Root Mean Squared Error for Relative Humidity is : ", rmse_humidity)

exp_var_score_temp = explained_variance_score(testl_array[:,0], DTpred[:,0])
print("The Explained variance score Temperature is {}".format(exp_var_score_temp))

exp_var_score_humidity = explained_variance_score(testl_array[:,1], DTpred[:,1])
print("The Explained variance score in terms of Relative Humidity is {}".format(exp_var_score_humidity))

mae_temp = mean_absolute_error(testl_array[:,0], DTpred[:,0])
print("Explained Variance Score for Temperature is ", mae_temp)

mae_hum = mean_absolute_error(testl_array[:,1], DTpred[:,1])
print("Explained Variance Score for Relative Humidity is ", mae_hum)

mean_abs_perc_error_temp = mean_absolute_percentage_error(testl_array[:,0], DTpred[:,0])
print("The mean absolute percentage error for Temperature is" ,mean_abs_perc_error_temp)

mean_abs_perc_error_humidity = mean_absolute_percentage_error(testl_array[:,1], DTpred[:,1])
print("The mean absolute percentage error for Relative Humidity is" ,mean_abs_perc_error_humidity)

Mean Squared Error for Temperature is :  0.32774279562206793
Mean Squared Error for Relative Humidity is :  1.5895936860567086
Root Mean Squared Error for Temperature is :  0.5724882493309954
Root Mean Squared Error for Relative Humidity is :  1.26079089703912
The Explained variance score Temperature is 0.9917347219801436
The Explained variance score in terms of Relative Humidity is 0.9950123379544541
Explained Variance Score for Temperature is  0.34247257538000586
Explained Variance Score for Relative Humidity is  0.6624524992692196
The mean absolute percentage error for Temperature is 3.1390522502086773
The mean absolute percentage error for Relative Humidity is 1.2082709443972117

ModelMetrics= ModelMetrics.append([['Decison Tree Regressor',mse_temp,mse_humidity,rmse_temp,rmse_humidity,exp_var_score_temp,exp_var_score_humidity,mae_temp,mae_hum,mean_abs_perc_error_temp,mean_abs_perc_error_humidity]])

ModelMetrics

	0	1	2	3	4	5	6	7	8	9	10
0	Linear Regression	0.21	23.61	0.45	4.86	0.99	0.93	0.33	3.35	3.30	5.46
0	KNN Regression	0.88	21.92	0.94	4.68	0.98	0.93	0.69	3.33	5.81	5.08
0	SVR-Multioutput	0.22	24.60	0.47	4.96	0.99	0.92	0.33	3.26	3.61	5.50
0	SVR-RegressorChain	0.22	24.64	0.47	4.96	0.99	0.92	0.33	3.27	3.58	5.49
0	Decison Tree Regressor	0.33	1.59	0.57	1.26	0.99	1.00	0.34	0.66	3.14	1.21

#Temperature Prediction
plt.plot(range(0,100), testl_array[0:100,0], label="Actual_Temp", color='r')
plt.scatter(range(0,100), DTpred[0:100,0], label="Predicted_Temp", color='b')
plt.xlabel('Index')
plt.ylabel('Temperature in Degrees Celcius')
plt.legend()
plt.show()

#RH Prediction
plt.plot(range(0,100), testl_array[0:100,1], label="Actual_Humidity", color='g')
plt.scatter(range(0,100), DTpred[0:100,1], label="Predicted_Humidity", color='y')
plt.xlabel('Index')
plt.ylabel('Percentage Humidity')
plt.legend()
plt.show()

#Temperature Scatter Plot for predictions
plt.scatter(testl_array[:,0], DTpred[:,0])
plt.ylabel("Predicted Temperatures")
plt.xlabel("Test Temperatures")
plt.title(" Predicted Vs. Test Temperature")
plt.show()

#RH Scatter Plot for predictions
plt.scatter(testl_array[:,1], DTpred[:,1])
plt.ylabel("Predicted RH")
plt.xlabel("Test RH")
plt.title(" Predicted Vs. Test RH")
plt.show()

Gradient Boosted Trees - Scikit Learn

gbr = GradientBoostingRegressor()
model_gbr = MultiOutputRegressor(estimator=gbr)
model_gbr.fit(trainf_array_scaled,trainl_array)

MultiOutputRegressor(estimator=GradientBoostingRegressor(alpha=0.9,
                                                         ccp_alpha=0.0,
                                                         criterion='friedman_mse',
                                                         init=None,
                                                         learning_rate=0.1,
                                                         loss='ls', max_depth=3,
                                                         max_features=None,
                                                         max_leaf_nodes=None,
                                                         min_impurity_decrease=0.0,
                                                         min_impurity_split=None,
                                                         min_samples_leaf=1,
                                                         min_samples_split=2,
                                                         min_weight_fraction_leaf=0.0,
                                                         n_estimators=100,
                                                         n_iter_no_change=None,
                                                         presort='deprecated',
                                                         random_state=None,
                                                         subsample=1.0,
                                                         tol=0.0001,
                                                         validation_fraction=0.1,
                                                         verbose=0,
                                                         warm_start=False),
                     n_jobs=None)

GBRpred=model_gbr.predict(testf_array_scaled)

#Model Evaluation

mse_temp = mean_squared_error(testl_array[:,0], GBRpred[:,0])
print("Mean Squared Error for Temperature is : ", mse_temp)

mse_humidity = mean_squared_error(testl_array[:,1], GBRpred[:,1])
print("Mean Squared Error for Relative Humidity is : ", mse_humidity)

rmse_temp = math.sqrt(mean_squared_error(testl_array[:,0], GBRpred[:,0]))
print("Root Mean Squared Error for Temperature is : ", rmse_temp)

rmse_humidity = math.sqrt(mean_squared_error(testl_array[:,1], GBRpred[:,1]))
print("Root Mean Squared Error for Relative Humidity is : ", rmse_humidity)

exp_var_score_temp = explained_variance_score(testl_array[:,0], GBRpred[:,0])
print("The Explained variance score Temperature is {}".format(exp_var_score_temp))

exp_var_score_humidity = explained_variance_score(testl_array[:,1], GBRpred[:,1])
print("The Explained variance score in terms of Relative Humidity is {}".format(exp_var_score_humidity))

mae_temp = mean_absolute_error(testl_array[:,0], GBRpred[:,0])
print("Explained Variance Score for Temperature is ", mae_temp)

mae_hum = mean_absolute_error(testl_array[:,1], GBRpred[:,1])
print("Explained Variance Score for Relative Humidity is ", mae_hum)

mean_abs_perc_error_temp = mean_absolute_percentage_error(testl_array[:,0], GBRpred[:,0])
print("The mean absolute percentage error for Temperature is" ,mean_abs_perc_error_temp)

mean_abs_perc_error_humidity = mean_absolute_percentage_error(testl_array[:,1], GBRpred[:,1])
print("The mean absolute percentage error for Relative Humidity is" ,mean_abs_perc_error_humidity)

Mean Squared Error for Temperature is :  0.17083562044173928
Mean Squared Error for Relative Humidity is :  2.934639318517773
Root Mean Squared Error for Temperature is :  0.4133226589986802
Root Mean Squared Error for Relative Humidity is :  1.7130789002605142
The Explained variance score Temperature is 0.9956924634782842
The Explained variance score in terms of Relative Humidity is 0.9908027240697022
Explained Variance Score for Temperature is  0.29705702368634024
Explained Variance Score for Relative Humidity is  1.1105368710712196
The mean absolute percentage error for Temperature is 2.9951625843664065
The mean absolute percentage error for Relative Humidity is 1.956731505537482

ModelMetrics= ModelMetrics.append([['Gradient Boosted Trees',mse_temp,mse_humidity,rmse_temp,rmse_humidity,exp_var_score_temp,exp_var_score_humidity,mae_temp,mae_hum,mean_abs_perc_error_temp,mean_abs_perc_error_humidity]])

ModelMetrics

	0	1	2	3	4	5	6	7	8	9	10
0	Linear Regression	0.21	23.61	0.45	4.86	0.99	0.93	0.33	3.35	3.30	5.46
0	KNN Regression	0.88	21.92	0.94	4.68	0.98	0.93	0.69	3.33	5.81	5.08
0	SVR-Multioutput	0.22	24.60	0.47	4.96	0.99	0.92	0.33	3.26	3.61	5.50
0	SVR-RegressorChain	0.22	24.64	0.47	4.96	0.99	0.92	0.33	3.27	3.58	5.49
0	Decison Tree Regressor	0.33	1.59	0.57	1.26	0.99	1.00	0.34	0.66	3.14	1.21
0	Gradient Boosted Trees	0.17	2.93	0.41	1.71	1.00	0.99	0.30	1.11	3.00	1.96

#Temperature Prediction
plt.plot(range(0,100), testl_array[0:100,0], label="Actual_Temp", color='r')
plt.scatter(range(0,100), GBRpred[0:100,0], label="Predicted_Temp", color='b')
plt.xlabel('Index')
plt.ylabel('Temperature in Degrees Celcius')
plt.legend()
plt.show()

#RH Prediction
plt.plot(range(0,100), testl_array[0:100,1], label="Actual_Humidity", color='g')
plt.scatter(range(0,100), GBRpred[0:100,1], label="Predicted_Humidity", color='y')
plt.xlabel('Index')
plt.ylabel('Percentage Humidity')
plt.legend()
plt.show()

#Temperature Scatter Plot for predictions
plt.scatter(testl_array[:,0], GBRpred[:,0])
plt.ylabel("Predicted Temperatures")
plt.xlabel("Test Temperatures")
plt.title(" Predicted Vs. Test Temperature")
plt.show()

#RH Scatter Plot for predictions
plt.scatter(testl_array[:,1], GBRpred[:,1])
plt.ylabel("Predicted RH")
plt.xlabel("Test RH")
plt.title(" Predicted Vs. Test RH")
plt.show()

Gradient Boostes Trees - LightGBM

model_lgb = MultiOutputRegressor(lgb.LGBMRegressor(random_state=41), n_jobs=-1)
model_lgb.fit(trainf_array_scaled, trainl_array)

MultiOutputRegressor(estimator=LGBMRegressor(boosting_type='gbdt',
                                             class_weight=None,
                                             colsample_bytree=1.0,
                                             importance_type='split',
                                             learning_rate=0.1, max_depth=-1,
                                             min_child_samples=20,
                                             min_child_weight=0.001,
                                             min_split_gain=0.0,
                                             n_estimators=100, n_jobs=-1,
                                             num_leaves=31, objective=None,
                                             random_state=41, reg_alpha=0.0,
                                             reg_lambda=0.0, silent=True,
                                             subsample=1.0,
                                             subsample_for_bin=200000,
                                             subsample_freq=0),
                     n_jobs=-1)

LGBpred = model_lgb.predict(testf_array_scaled)

#Model Evaluation

mse_temp = mean_squared_error(testl_array[:,0], LGBpred[:,0])
print("Mean Squared Error for Temperature is : ", mse_temp)

mse_humidity = mean_squared_error(testl_array[:,1], LGBpred[:,1])
print("Mean Squared Error for Relative Humidity is : ", mse_humidity)

rmse_temp = math.sqrt(mean_squared_error(testl_array[:,0], LGBpred[:,0]))
print("Root Mean Squared Error for Temperature is : ", rmse_temp)

rmse_humidity = math.sqrt(mean_squared_error(testl_array[:,1], LGBpred[:,1]))
print("Root Mean Squared Error for Relative Humidity is : ", rmse_humidity)

exp_var_score_temp = explained_variance_score(testl_array[:,0], LGBpred[:,0])
print("The Explained variance score Temperature is {}".format(exp_var_score_temp))

exp_var_score_humidity = explained_variance_score(testl_array[:,1], LGBpred[:,1])
print("The Explained variance score in terms of Relative Humidity is {}".format(exp_var_score_humidity))

mae_temp = mean_absolute_error(testl_array[:,0], LGBpred[:,0])
print("Explained Variance Score for Temperature is ", mae_temp)

mae_hum = mean_absolute_error(testl_array[:,1], LGBpred[:,1])
print("Explained Variance Score for Relative Humidity is ", mae_hum)

mean_abs_perc_error_temp = mean_absolute_percentage_error(testl_array[:,0], LGBpred[:,0])
print("The mean absolute percentage error for Temperature is" ,mean_abs_perc_error_temp)

mean_abs_perc_error_humidity = mean_absolute_percentage_error(testl_array[:,1], LGBpred[:,1])
print("The mean absolute percentage error for Relative Humidity is" ,mean_abs_perc_error_humidity)

Mean Squared Error for Temperature is :  0.07829868794607751
Mean Squared Error for Relative Humidity is :  0.8078769280297866
Root Mean Squared Error for Temperature is :  0.27981902713374857
Root Mean Squared Error for Relative Humidity is :  0.898819741677822
The Explained variance score Temperature is 0.9980254235822068
The Explained variance score in terms of Relative Humidity is 0.9974675512827723
Explained Variance Score for Temperature is  0.1918676768015798
Explained Variance Score for Relative Humidity is  0.5961751869917571
The mean absolute percentage error for Temperature is 2.73355517934084
The mean absolute percentage error for Relative Humidity is 1.0256503272653266

ModelMetrics= ModelMetrics.append([['Gradient Boosted Trees - LGB',mse_temp,mse_humidity,rmse_temp,rmse_humidity,exp_var_score_temp,exp_var_score_humidity,mae_temp,mae_hum,mean_abs_perc_error_temp,mean_abs_perc_error_humidity]])

ModelMetrics

	0	1	2	3	4	5	6	7	8	9	10
0	Linear Regression	0.21	23.61	0.45	4.86	0.99	0.93	0.33	3.35	3.30	5.46
0	KNN Regression	0.88	21.92	0.94	4.68	0.98	0.93	0.69	3.33	5.81	5.08
0	SVR-Multioutput	0.22	24.60	0.47	4.96	0.99	0.92	0.33	3.26	3.61	5.50
0	SVR-RegressorChain	0.22	24.64	0.47	4.96	0.99	0.92	0.33	3.27	3.58	5.49
0	Decison Tree Regressor	0.33	1.59	0.57	1.26	0.99	1.00	0.34	0.66	3.14	1.21
0	Gradient Boosted Trees	0.17	2.93	0.41	1.71	1.00	0.99	0.30	1.11	3.00	1.96
0	Gradient Boosted Trees - LGB	0.08	0.81	0.28	0.90	1.00	1.00	0.19	0.60	2.73	1.03

#Temperature Prediction
plt.plot(range(0,100), testl_array[0:100,0], label="Actual_Temp", color='r')
plt.scatter(range(0,100), LGBpred[0:100,0], label="Predicted_Temp", color='b')
plt.xlabel('Index')
plt.ylabel('Temperature in Degrees Celcius')
plt.legend()
plt.show()

#RH Prediction
plt.plot(range(0,100), testl_array[0:100,1], label="Actual_Humidity", color='g')
plt.scatter(range(0,100), LGBpred[0:100,1], label="Predicted_Humidity", color='y')
plt.xlabel('Index')
plt.ylabel('Percentage Humidity')
plt.legend()
plt.show()

#Temperature Scatter Plot for predictions
plt.scatter(testl_array[:,0], LGBpred[:,0])
plt.ylabel("Predicted Temperatures")
plt.xlabel("Test Temperatures")
plt.title(" Predicted Vs. Test Temperature")
plt.show()

#RH Scatter Plot for predictions
plt.scatter(testl_array[:,1], LGBpred[:,1])
plt.ylabel("Predicted RH")
plt.xlabel("Test RH")
plt.title(" Predicted Vs. Test RH")
plt.show()

Random Forest Model

#model = RandomForestRegressor(n_estimators=100, bootstrap='TRUE', max_depth=20)
model_RF = RandomForestRegressor(n_jobs=-1, random_state=35)
model_RF.fit(trainf_array,trainl_array)


#parameters used by the current random forest model initiated. 
from pprint import pprint
print('Parameters currently in use:\n')
pprint(model.get_params())

Parameters currently in use:

{'bootstrap': True,
 'ccp_alpha': 0.0,
 'criterion': 'mse',
 'max_depth': None,
 'max_features': 'auto',
 'max_leaf_nodes': None,
 'max_samples': None,
 'min_impurity_decrease': 0.0,
 'min_impurity_split': None,
 'min_samples_leaf': 1,
 'min_samples_split': 2,
 'min_weight_fraction_leaf': 0.0,
 'n_estimators': 100,
 'n_jobs': None,
 'oob_score': False,
 'random_state': 42,
 'verbose': 0,
 'warm_start': False}

RFpred = model_RF.predict(testf_array)

#Model Evaluation

mse_temp = mean_squared_error(testl_array[:,0], RFpred[:,0])
print("Mean Squared Error for Temperature is : ", mse_temp)

mse_humidity = mean_squared_error(testl_array[:,1], RFpred[:,1])
print("Mean Squared Error for Relative Humidity is : ", mse_humidity)

rmse_temp = math.sqrt(mean_squared_error(testl_array[:,0], RFpred[:,0]))
print("Root Mean Squared Error for Temperature is : ", rmse_temp)

rmse_humidity = math.sqrt(mean_squared_error(testl_array[:,1], RFpred[:,1]))
print("Root Mean Squared Error for Relative Humidity is : ", rmse_humidity)

exp_var_score_temp = explained_variance_score(testl_array[:,0], RFpred[:,0])
print("The Explained variance score Temperature is {}".format(exp_var_score_temp))

exp_var_score_humidity = explained_variance_score(testl_array[:,1], RFpred[:,1])
print("The Explained variance score in terms of Relative Humidity is {}".format(exp_var_score_humidity))

mae_temp = mean_absolute_error(testl_array[:,0], RFpred[:,0])
print("Explained Variance Score for Temperature is ", mae_temp)

mae_hum = mean_absolute_error(testl_array[:,1], RFpred[:,1])
print("Explained Variance Score for Relative Humidity is ", mae_hum)

mean_abs_perc_error_temp = mean_absolute_percentage_error(testl_array[:,0], RFpred[:,0])
print("The mean absolute percentage error for Temperature is" ,mean_abs_perc_error_temp)

mean_abs_perc_error_humidity = mean_absolute_percentage_error(testl_array[:,1], RFpred[:,1])
print("The mean absolute percentage error for Relative Humidity is" ,mean_abs_perc_error_humidity)

Mean Squared Error for Temperature is :  0.106908996407001
Mean Squared Error for Relative Humidity is :  0.4812867436422099
Root Mean Squared Error for Temperature is :  0.3269694120357453
Root Mean Squared Error for Relative Humidity is :  0.6937483287491293
The Explained variance score Temperature is 0.9973046016152061
The Explained variance score in terms of Relative Humidity is 0.9984904946613009
Explained Variance Score for Temperature is  0.1992759622673287
Explained Variance Score for Relative Humidity is  0.3816113709441685
The mean absolute percentage error for Temperature is 2.0283611717550016
The mean absolute percentage error for Relative Humidity is 0.6935920178001578

ModelMetrics= ModelMetrics.append([['Random Forest Regression',mse_temp,mse_humidity,rmse_temp,rmse_humidity,exp_var_score_temp,exp_var_score_humidity,mae_temp,mae_hum,mean_abs_perc_error_temp,mean_abs_perc_error_humidity]])

ModelMetrics

	0	1	2	3	4	5	6	7	8	9	10
0	Linear Regression	0.21	23.61	0.45	4.86	0.99	0.93	0.33	3.35	3.30	5.46
0	KNN Regression	0.88	21.92	0.94	4.68	0.98	0.93	0.69	3.33	5.81	5.08
0	SVR-Multioutput	0.22	24.60	0.47	4.96	0.99	0.92	0.33	3.26	3.61	5.50
0	SVR-RegressorChain	0.22	24.64	0.47	4.96	0.99	0.92	0.33	3.27	3.58	5.49
0	Decison Tree Regressor	0.33	1.59	0.57	1.26	0.99	1.00	0.34	0.66	3.14	1.21
0	Gradient Boosted Trees	0.17	2.93	0.41	1.71	1.00	0.99	0.30	1.11	3.00	1.96
0	Gradient Boosted Trees - LGB	0.08	0.81	0.28	0.90	1.00	1.00	0.19	0.60	2.73	1.03
0	Random Forest Regression	0.11	0.48	0.33	0.69	1.00	1.00	0.20	0.38	2.03	0.69

#Temperature Prediction
plt.plot(range(0,100), testl_array[0:100,0], label="Actual_Temp", color='r')
plt.scatter(range(0,100), RFpred[0:100,0], label="Predicted_Temp", color='b')
plt.xlabel('Index')
plt.ylabel('Temperature in Degrees Celcius')
plt.legend()
plt.show()

#RH Prediction
plt.plot(range(0,100), testl_array[0:100,1], label="Actual_Humidity", color='g')
plt.scatter(range(0,100), RFpred[0:100,1], label="Predicted_Humidity", color='y')
plt.xlabel('Index')
plt.ylabel('Percentage Humidity')
plt.legend()
plt.show()

#Temperature Scatter Plot for predictions
plt.scatter(testl_array[:,0], RFpred[:,0])
plt.ylabel("Predicted Temperatures")
plt.xlabel("Test Temperatures")
plt.title(" Predicted Vs. Test Temperature")
plt.show()

#RH Scatter Plot for predictions
plt.scatter(testl_array[:,1], RFpred[:,1])
plt.ylabel("Predicted RH")
plt.xlabel("Test RH")
plt.title(" Predicted Vs. Test RH")
plt.show()

Overall Predictions

Final_Predictions = pd.DataFrame({'Test_Temp':testl_array[:,0],'Test_RH':testl_array[:,1],'Linear_Reg_temp':LRpred[:,0],'Linear_Reg_RH':LRpred[:,1], 'KNN_Reg_temp':KNNpred[:,0], 'KNN_Reg_RH':KNNpred[:,1], 'SVR_temp':SVRpred[:,0], 'SVR_RH':SVRpred[:,1], 'RCSVR_temp':RCSVRpred[:,0], 'RCSVR_RH':RCSVRpred[:,1], 'DecisionTree_temp':DTpred[:,0], 'DecisionTree_RH': DTpred[:,1], 'GradentBoostedTree_temp':GBRpred[:,0], 'GradentBoostedTree_RH':GBRpred[:,1], 'LGB_temp':LGBpred[:,0], 'LGB_RH':LGBpred[:,1], 'RandomForest_temp':RFpred[:,0], 'RandomForest_RH':RFpred[:,1]})

Final_Predictions.head(25)

	Test_Temp	Test_RH	Linear_Reg_temp	Linear_Reg_RH	KNN_Reg_temp	KNN_Reg_RH	SVR_temp	SVR_RH	RCSVR_temp	RCSVR_RH	DecisionTree_temp	DecisionTree_RH	GradentBoostedTree_temp	GradentBoostedTree_RH	LGB_temp	LGB_RH	RandomForest_temp	RandomForest_RH
0	21.74	73.0	21.65	74.67	23.55	65.6	21.57	75.45	21.61	75.25	23.32	74.0	21.73	75.37	22.22	73.32	21.79	73.19
1	22.88	67.0	22.61	66.78	23.24	66.8	22.79	67.22	22.78	66.80	22.99	68.0	22.83	67.56	23.04	67.52	22.90	67.04
2	29.25	56.0	28.98	53.59	29.05	61.6	28.91	53.65	28.92	53.63	29.71	57.0	29.31	58.33	29.13	57.08	29.62	56.92
3	30.29	60.0	30.60	59.82	29.80	59.6	30.69	59.35	30.66	59.05	29.49	58.0	30.25	58.97	30.27	59.21	29.91	59.15
4	17.63	57.0	16.98	55.38	15.80	45.0	17.10	58.19	17.12	57.76	17.37	56.0	17.42	51.31	17.18	54.83	17.60	56.58
5	24.77	65.0	24.34	68.56	23.64	66.2	24.46	68.39	24.49	67.99	24.86	66.0	25.08	60.40	24.57	64.16	24.95	65.51
6	15.00	72.0	15.02	76.74	17.05	75.6	15.04	77.08	14.99	77.08	14.73	72.0	15.66	74.97	15.24	72.39	15.32	72.43
7	12.94	81.0	13.03	78.72	13.67	76.8	13.04	77.67	12.99	77.67	13.72	82.0	13.17	82.47	13.13	80.98	12.96	80.96
8	13.22	72.0	12.97	74.25	14.06	65.0	13.10	74.42	13.07	74.81	13.03	72.0	12.57	69.97	12.75	71.30	13.03	71.60
9	24.06	77.0	23.89	78.07	24.38	75.6	23.92	77.71	23.92	77.36	23.19	76.0	23.80	77.20	24.01	77.33	24.05	77.02
10	27.43	69.0	27.43	69.93	27.20	69.8	27.41	69.73	27.43	69.58	27.42	69.0	27.37	69.64	27.51	69.70	27.47	69.10
11	28.65	61.0	28.06	63.00	28.73	61.0	28.22	62.13	28.16	61.88	28.70	62.0	28.30	60.16	28.61	61.78	28.56	60.83
12	24.61	81.0	24.55	80.65	25.00	82.8	24.61	80.71	24.58	80.39	24.72	81.0	24.74	81.62	24.70	81.70	24.81	81.25
13	28.01	48.0	27.72	49.13	27.88	50.0	27.79	48.64	27.73	48.83	27.32	46.0	28.01	46.85	27.79	48.16	27.87	47.84
14	25.73	93.0	26.02	91.66	26.10	89.0	25.97	91.29	25.90	91.38	25.73	93.0	25.45	92.67	25.62	92.72	25.47	92.87
15	29.85	60.0	29.93	57.58	29.56	60.8	29.91	57.13	29.93	57.34	29.83	60.0	29.93	59.01	29.67	60.14	29.75	59.71
16	26.79	45.0	26.47	44.69	26.00	47.6	26.50	43.30	26.49	43.67	27.69	47.0	26.50	48.04	26.68	45.65	27.01	45.65
17	21.76	70.0	21.74	69.33	22.73	70.0	21.81	70.17	21.77	70.02	22.41	72.0	21.69	70.31	21.51	70.10	21.52	69.53
18	21.81	92.0	22.65	92.85	22.50	87.8	22.65	91.36	22.64	91.21	21.86	92.0	22.06	91.61	22.17	91.82	22.03	92.11
19	20.09	52.0	19.82	57.04	19.11	54.8	19.94	58.52	19.92	58.19	20.34	50.0	19.89	50.75	20.12	52.57	20.26	52.14
20	34.99	45.0	34.94	37.20	33.82	45.8	34.83	36.10	34.82	36.45	34.80	47.0	34.73	44.62	35.01	45.86	34.64	44.32
21	28.44	66.0	28.45	66.98	28.88	63.0	28.48	66.92	28.48	66.72	28.44	66.0	28.25	66.48	28.32	65.90	28.33	65.81
22	23.62	69.0	23.42	71.52	22.07	80.6	23.47	71.25	23.45	71.07	24.55	70.0	24.13	65.85	23.58	69.45	23.71	69.12
23	8.88	98.0	8.58	93.37	9.67	95.6	8.75	93.09	8.71	93.27	8.85	98.0	8.51	96.41	9.05	97.17	9.05	98.19
24	21.29	78.0	21.17	81.89	21.40	81.0	21.24	80.80	21.25	80.62	21.32	78.0	21.13	79.14	21.36	77.70	21.43	78.11

Analysis of Model Performance

#Assign suitable columns to the dataframe that contains the metrics for the applied algorithms
ModelMetrics.columns= ['ModelName','MSE-Temp','MSE-RH','RMSE-Temp','RMSE-RH','R2Score-Temp','R2Score-RH','MAE-Temp','MAE-RH','MAPE-Temp','MAPE-RH',]

ModelMetrics

	ModelName	MSE-Temp	MSE-RH	RMSE-Temp	RMSE-RH	R2Score-Temp	R2Score-RH	MAE-Temp	MAE-RH	MAPE-Temp	MAPE-RH
0	Linear Regression	0.21	23.61	0.45	4.86	0.99	0.93	0.33	3.35	3.30	5.46
0	KNN Regression	0.88	21.92	0.94	4.68	0.98	0.93	0.69	3.33	5.81	5.08
0	SVR-Multioutput	0.22	24.60	0.47	4.96	0.99	0.92	0.33	3.26	3.61	5.50
0	SVR-RegressorChain	0.22	24.64	0.47	4.96	0.99	0.92	0.33	3.27	3.58	5.49
0	Decison Tree Regressor	0.33	1.59	0.57	1.26	0.99	1.00	0.34	0.66	3.14	1.21
0	Gradient Boosted Trees	0.17	2.93	0.41	1.71	1.00	0.99	0.30	1.11	3.00	1.96
0	Gradient Boosted Trees - LGB	0.08	0.81	0.28	0.90	1.00	1.00	0.19	0.60	2.73	1.03
0	Random Forest Regression	0.11	0.48	0.33	0.69	1.00	1.00	0.20	0.38	2.03	0.69

Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) are considered as the suitable error metrics for this analysis.

plt.barh(ModelMetrics['ModelName'], ModelMetrics['MAE-RH'], color ='green')
plt.xlabel('RH Prediction Error (in % RH)')
plt.title('RH Prediction MAE for Different Models')
plt.show()

plt.barh(ModelMetrics['ModelName'], ModelMetrics['MAE-Temp'], color ='red')
plt.xlabel('Temperature Prediction Error (in Degrees Celcius)')
plt.title('Temperature Prediction MAE for Different Models')
plt.show()

plt.barh(ModelMetrics['ModelName'], ModelMetrics['MAPE-Temp'], color ='purple')
plt.xlabel('MAPE in Temperature Prediction')
plt.title('Temperature Prediction MAPE for Different Models')
plt.show()

plt.barh(ModelMetrics['ModelName'], ModelMetrics['MAPE-RH'], color ='brown')
plt.xlabel('MAPE in RH Prediction')
plt.title('RH Prediction MAPE for Different Models')
plt.show()